Nonfunctional requirements describe how a system should work under the hood rather than what features it provides. They cover qualities and constraints such as performance, security, portability, legal obligations, and privacy, making them central to responsible ML system design. section 2.3 section 13.3
Treat nonfunctional requirements as concrete questions about system risk and operation:
- Define what must be protected, identify the worst plausible scenarios, and align the answers with all relevant stakeholders. section 2.3
- Extend design beyond feature integration to include the technical and nontechnical connections needed for maintenance, product operations, legal compliance, and privacy. section 13.3
- Replace vague risks such as “incorrect model prediction” with specific consequences and protections. section 2.3
Dynamic pricing
A real-time dynamic pricing system combined revenue, margin, or traffic objectives with constraints, rapid adaptation, uncertainty, and user-level operation. Its design illustrates why functional value must be considered alongside performance and operational risk. section 2.3
